| { | |
| "architectures": [ | |
| "Ernie4_5_MoeModel" | |
| ], | |
| "bos_token_id": 1, | |
| "dtype": "float32", | |
| "eos_token_id": 2, | |
| "hidden_act": "gelu", | |
| "hidden_size": 32, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 32, | |
| "max_position_embeddings": 512, | |
| "model_type": "ernie4_5_moe", | |
| "moe_intermediate_size": 16, | |
| "moe_k": 2, | |
| "moe_layer_end_index": 1, | |
| "moe_layer_interval": 1, | |
| "moe_layer_start_index": 0, | |
| "moe_norm_min": 1e-12, | |
| "moe_num_experts": 8, | |
| "moe_num_shared_experts": 2, | |
| "num_attention_heads": 2, | |
| "num_hidden_layers": 2, | |
| "num_key_value_heads": 2, | |
| "output_router_logits": false, | |
| "pad_token_id": 0, | |
| "rms_norm_eps": 1e-05, | |
| "rope_parameters": { | |
| "rope_theta": 500000.0, | |
| "rope_type": "default" | |
| }, | |
| "router_aux_loss_coef": 0.001, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "5.16.0.dev0", | |
| "use_bias": false, | |
| "use_cache": true, | |
| "vocab_size": 101304 | |
| } | |